<![CDATA[This study addresses the challenge of incorporatingenvironmental data, particularly air pollution, into practicalactuarial models. This study extends the Lee-Carter modelusing environmental variables and employs LASSO regressionto address potential multicollinearity among pollution indicators.The results are closed-form approximations for estimating theexpected changes in transition probabilities from the previoustransition probabilities forecast within both multiple-state andmultiple-decrement models, which are calculated using currentenvironmental data. These approximations enable actuaries toquantify the expected difference in transition probabilities fromthe initial forecast by utilizing the newest environmental data thatwas previously unavailable. The framework is empirically testedusing a case study on a dementia-related multiple-state modelin the UK. Results show significant improvements in predictiveaccuracy, measured by reductions in Mean Absolute PercentageError (MAPE), reducing MAPE by over 40% in several agegroups. These results emphasize the value of environmentalintegration in actuarial modeling, particularly in the context ofhealth-related insurance products]]>
Copyrights © 2026
